Investigating the effect of Ser256 phosphorylation on gating of aquaporin-2: Molecular Dynamics study

Regulation of water transport via aquaporins is crucial for osmoregulation and water homeostasis of an organism. This transport of water is regulated either by gating or trafficking wherein AQPs are transported from intracellular storage sites to plasma membrane. It has been proposed that water movement via AQP2 is regulated by post-translational modification. We aimed to explore the structural and functional changes occurring in AQP2 due to Ser256 phosphorylation. We have carried out molecular dynamics simulations to investigate molecular basis of effect of phosphorylation on water permeability of AQP2. MD simulations show that there are mild variations in the pore sizes of different monomers of the phosphorylated and unphosphorylated AQP2. Analysis of inter and intra-monomeric interactions such as hydrogen bond, electrostatic and hydrophobic interactions has been carried out. Structures of the phosphorylated AQP2 do not show any blocking of mouth of pore of the monomers during the course of MD simulations. Further, water permeability calculations do corroborate the above finding. This molecular dynamics study suggests that phosphorylation of C-terminal Ser-256 residue of AQP2 may not be directly responsible for gating mechanism.

Cs+, Na+,and tetramethylammonium (Saparov et al., 2001;Yool and Weinstein, 2002). All 10 aquaporins have conserved NPA (Asn-Pro-Ala) motif near the center and aromatic/Arginine 11 (ar/R) selectivity filter towards extracellular side in common and these two form constriction 12 regions within the channel. AQPs participate in many physiological and pathophysiological 13 processes that include renal water absorption, brain water homeostasis, fat metabolism, liver 14 gluconeogenesis, tumor angiogenesis and reproduction (Soveral and Casini, 2016). Regulation  at residue Ser and Thr facing cytoplasmic region. This leads to de-protonation of His183 22 residue facing outer membrane which leads to widening of the channel pore. Whereas at pH 23 5.1, the protonated residue and putative hydrogen bond interaction keeps the channel pore 24 narrow but open resulting in lowering of the water permeability. There are numerous evidences 25 of regulation of aquaporins, majority of these uses gating mechanism affecting the rate of water 1 permeability dependent on phosphorylation. However, the possibility of AQP2 gating 2 mechanism has not been rigorously explored.

3
AQP2 has been reported to be regulated by vasopressin-induced phosphorylation causing its 4 trafficking to the apical membrane. AQP2 contains many putative kinase recognition residues 5 such as Ser256, Ser261, Ser264, Ser269, and these C-terminal residues that are suggested to 6 play important role in trafficking of AQP2 (Moeller et al., 2009a). Phosphorylation of Ser256 7 has been suggested to participate in AQP2 translocation to apical membrane but doesn't affect 8 water permeability through AQP2 channel (Brown et al., 2008). Phosphorylated AQP2 has 9 comparable water permeability with that of wild type AQP2 whereas dephosphorylation of 10 AQP2 reports threefold reduction in osmotic water permeability in comparison of wild type    Material and method 21 The unphosphorylated X-Ray diffracted crystal structure of` human aquaporin-2 with 22 resolution 3.05Å, has been obtained from the Protein Data Bank (PDB ID: 4OJ2) (Vahedi- 23 Faridi et al., 2013). 3D structure was monomeric; thus, tetrameric form was constructed using 24 online serverMakeMultimer.py (http://watcut.uwaterloo.ca/tools/makemultimer/index). 1 Crystal structure contains engineered amino acid at Ala 256 which was substituted with Ser256 2 identical to wildtype hAQP2. We were interested in studying effect of phosphorylation of 3 Ser256 on water permeability. Therefore, phosphorylated form of AQP2 was generated by 4 adding a phosphate group to Ser256 residue using CHARMM patch in NAMD (Phillips et al., of lipid bilayer while exposed part of AQP2 was surrounded by a layer of 20Å thick water 10 molecules. Total number of water molecules in phosphorylated and unphosphorylated system 11 was 28546 and 28502, respectively. Net charge of the system was made zero by adding 12 appropriate number of counter ions using AUTOIONIZE plugin of VMD and ionic 13 concentration was set to 150mM by adding Na + and Clions. 14 MD simulations 15 Phosphorylated as well as the unphosphorylated structure of AQP2 were minimized in three 16 stages Initially, both system were minimized for 8000 steps of steepest descent and then for 17 15000 steps of conjugate gradient. In third stage, system was minimized for 20000 conjugate 18 gradient steps keeping the ar/R site residues fixed. Only lipid tails of minimized system were 19 allowed to melt at 450K in NVT ensemble for 200 ps, while keeping rest of atoms (water, lipid 20 head group, ions, and protein) fixed in order to introduce appropriate disorder of fluid like 21 bilayer. The first equilibration was performed for 300ps at 310K by applying constraint of 4, 22 3, 2 and 5 kcal/mol/Å on lipid headgroups, tails, water, and protein, respectively. Later on, the 23 system was equilibrated for 1000ps by reducing the constraint to 3, 2, 1 and 4 kcal/mol/Å. 24 Again a constrained equilibration of 1000ps was performed applying force constant of 2, 1, 0, molecules. An additional constraint of 4 kcal/mol/Å has been applied on dihedral angle of 4 Arginine (Arg) residue of ar/R site to prevent initial protrusion and hence blocking of the pore.

5
Further equilibration was carried out at 310K in NPT ensemble fixing only the protein atoms 6 and dihedral angles of Arg with the constraint of 2 and 4 kcal/mol/Å, respectively for 3.5ns.  The single-file water molecules permeating through a channel was quantified using diffusion 23 permeability coefficient (p d ) given by 24 25 where v w is the volume of a single water molecule and q 0 is the number of complete permeation 2 events in one direction across the channel in unit time (Zhu et al., 2004). If the average number 3 of water molecules in the lumen of the channel is N, the osmotic permeability constant (p f ) can 4 be calculated as We have calculated the rate of diffusion permeability and osmotic permeability coefficient for  High RMSDs of C-terminal region possibly contribute to the high RMSD of the tetramer 23 assembly of AQP2. Despite the overall high RMSD, the C-terminal region keeps its 24 compactness which has been confirmed by calculations of Radius of Gyration (R gyr ) as shown 25 in (Fig 2a,   represented by black and grey colours, respectively for each monomer of AQP2.

18
We have carried out disorder analysis of protein sequence using PrDOS (Ishida and Kinoshita,

19
2007) and disorder probability of each residue of AQP2 has been calculated. The prediction 20 results showed that C-terminal region residues of AQP2 have high disorder propensity in 21 comparison to rest of the protein (Fig 4a). High RMSFs in all the monomers of 22 unphosphorylated as well as phosphorylated AQP2 are mainly attributed to the movement of 23 C-terminal region present in the cytoplasmic side (Fig 4b).   interaction between side chain oxygen of Glu250 and side chain nitrogen of Arg253 found in 6 all three monomers except in monomer D with varying propensity of occurrence whereas, in 7 phosphorylated AQP2, the interactions between these residues were absent in monomer C and side chain nitrogen of Arg5 of N-terminal of monomer B (Fig 5a). Also, there is interaction Glu232 of monomer A is found in unphosphorylated form (Fig 5d). The latter interaction has 16 not been observed in phosphorylated form of AQP2. A few inter-monomeric interactions with 17 lower propensities were also observed and summarized in (Supplementay Table 2

22
Electrostatic potential surface analysis: 23 Snapshots of MD trajectories were extracted at an interval of 5ns from 100ns long MD 24 trajectory. The electrostatic potential surface was analyzed using Pymol (Schrodinger, 2015). 25 Surface representation of extracted structure at the start of MD simulations and at 25, 50, 75, 100ns of MD trajectory show that C-terminal is not able to cover the mouth of water channel 14 in any of the forms (Fig 6 a, b, c, and d). Similar behaviour of conformations of AQP2 were 15 found throughout the MD simulations. AQP2 are shown in cartoon representation.

20
Hole Analysis: 21 The pore radius of phosphorylated and unphosphorylated forms of AQP2 has been calculated 22 during MD simulations using HOLE program (Smart et al., 1996). Analysis shows that there 23 was a marginal difference in pore radius of two forms of AQP2 during MD trajectory. More  showing changes in pore radius across the channel coordinate at 5ns,25ns,50ns,75ns, and 100ns MD 14 simulations.

15
Effect of phosphorylation on water conductivity 16 The calculated diffusion permeability of unphosphorylated and phosphorylated AQP2 was  (Fig 8a and b).      India.